CTP Ratings (29-APR-2001)

[jpww]

quote: Originally posted by TheBirdMan on 05-01-2001 02:33 PM Or more easily - put them where they belong (as you do now)! But give them an * Ãn front of their rating, showing that this position is only "theoretic" until their first official rating...

I agree with the Birdman on this!

[blackice]

Hummm I am a bit discouraged here Quinns that was just a very small reason.
I hope no gets this wrong it is a bit messy the way things are done. It simply does not make much sense and did not when I signed up. I took a .05 loss necause the game did not start in the time frame. Like so what you can beat deity, the difference between this PBEM and online gaming is huge. You simply can not play the same way. But there are several reasons and that is a small one.

I also had said to Quinns and Franses (Thanks Franses) I did not want to post my reasons as they were of course personal and did not want to ruffle feathers, or discourage people from playing this great way PBEM. It is a riot and I would encourage anyone to play and have as much fun as I am having.

As I said I have 17 games on the go, obviously I like it heh. I forgive but am a bit disapointed that my desire to keep it to myself has been posted. More importantly only part of the reason was posted giving in my opinion the wrong impression for the reasons. But moving on cuz I have a game to play, more importantly the style of play is sooo different from online civ. big time. I just feel that my objective in one ladder is being affected by the other style of play. For you that can relate it's like playing darts, baseball and bowling all at the same time heh.

Here is my senario at turn say 40 or so my strategy kicks in ok I am fist over all but for four turns I have taken a bath in the ratings as I have been beaten by lower ranked players. The points ratio I will recieve by beating lower ranked players will never match the points I lost to begin with. In GL play you have the choice of who to play. Obviously you don't play lower ranked people, believe me I know heh 12th to 29th in five of games unless of course your name is Zeratul lol and even he can be sellective of his win losss points ratio. So the choice is yours who you play for points. That gives you control on how much you win or lose and how many points it is going to be. Here no, most players are as Quinns points out new comers, so it is obvious that these baths per ten turns will be endless especially at the starting rank. But the big difference is how much of a bath you take and how much of a win you take.

To me projecting into the future your loses will continually exceed your wins, points wise with my style of play. This was made clear to me with the last rankings, I had two first place finishes and one forth, the forth place finish not only wiped out the two first places but took more points than both wins were worth. I lost points? Ok so if I stay on top for two of them and forth place in the other I will be at the bottom of the rankings in two months lol. Personaly that simply makes no sense, at all.
If I continue to be first those players will continue to be lower ranked.
Therefore in each game I play where I win I have less points than the turns I lose, heh.

Unless they play other games and I hit bottom and they move above me then of course I will gain points until they get below me again by my wins over them and by the loss you take vs the gains you make that won't take long long. Heh see what I mean? Anyone? lol Feel free to correct me if I am wrong.


Thanks TheBirdMan I always wondered if I did your top notch style of play (as I know I played your civ.) justice. :0)


Quinns that is a great idea have a waiting to be ranked area like newbies ladder Heh let the games begin :0)

Blackice


[This message has been edited by blackice (edited May 01, 2001).]

[jpww]

Getting beaten by an unrated player doesn't count against you, does it?

[quinns]

Blackice was discouraged by the new players jumping ahead of most everyone in the ratings. I explained to him that the new players win/loss records indicate the "weight" of their ratings, but he was still put off by it. I can understand that. He has a good point. This has surfaced before.

New Proposed Rule

How about this PROVISIONAL rule then:

New players will not show up on the rankings reports until they have had at least 10 Power Graph results. Players with less than 10 Power Graph results will be considered PROVISIONAL and their results will be recorded just as normal, however they will not show in the rankings.

That is, a combination of at least 10 wins and/or losses. I think this is a very fair resolution to this distortion, and prevents players jumping to the top of the rankings without playing a single game. What do you all think?

[Franses]

Blackice makes some good points here but there is something to say for the current system also. I suggest a double ranking system for a while. The second ranking system I propose probably solves the issues raised by Blackice. After a couple of months both ranking systems should come close to eachother and if not we could decide on which one to use. My proposal for the second ranking system is a system simply based on wins/losses. I calculated it for the first 10 with the following result:

1. Stavros +42
2. Klair +27
Mobius +27
Blackice +27
5. Kral-M +25
6. Birdman +23
7. Berxpert +15
8. Swissy +13
9. Elkins +11
10.Slamp +10

Furthermore:
The game is all about winning it. Intermediate results are not that important, especially not because the Power Graph we use for the ranking does not reveal the true strength of a civilization. Yet, games take so long that an intermediate ranking system (or two ) like we have is worthwhile. Yet, I would also appreciate to see a ranking based on position at the end of the game. Why not have a third ranking that shows this too based e.g. on the following:

For each game ended the winner gets 10 points, second place 5 points (if at least 4 players took part only), third place 3 points (if at least 5 players took part only). The number of points is divided by the number of games played. So, if you ended three games with the positions 1, 3 and 5 you would have (10+3+0)/3=4.33 points.

[TheBirdMan]

Frans!

Personally I like your last proposal most. Wonder how our ranking would look like if using this?!

I'll make a calculation later today or tomorrow morning.

Right now the sun is shining (lovely) - and I had better stop playing and go out helping my wife with her customers (selling all kind (nearly) of flowers and trees) .

[Franses]

Birdie, buy a portable with a wireless connection and you can do both .

[Keygen]

I still suggest that we should start a new topic about these problems. There might be many things to discuss. It would be better to have them all together in one relative topic .

[This message has been edited by Keygen (edited May 02, 2001).]

[berXpert]

I'm agree with Blackice...


...that's all

Edit: I said nothing, quinns have allways the reason. In fact this is HIS rating system, if we don't like it then just go as BlackIce, as the same quinns sugges in the next post.

Me.... let me think about it!

[This message has been edited by berXpert (edited May 02, 2001).]

[quinns]

Not again! More challenges to the rating system. Where is the PBEM Ladder Administrator Solver when I need him? Okay -- here goes the defense.... AGAIN!

1. Blackice is incorrect about his conclusions regarding unfairly dropping in ratings. I could go on and on with example after example, but he is just mistaken. Trust me.

2. Basing ratings on the number of wins/losses will not work. Take the example of a player with average ability, playing many games against ONLY beginners. This average player would skyrocket to the top due to his enormous number of wins.

3. We already have a system in place now, that rewards players for winning in the end, (as well as punishing "quitters" as berXpert suggested to "cheat" the system). See the "Elimination Rule" above.

4. The current ratings system is now in place, functioning accurately, and is popular with most players. Do you really want to change it (even slightly) and risk losing many participants?


[This message has been edited by quinns (edited May 02, 2001).]

[Franses]

I, for one, did not suggest another rating system. Just a view from another angle TOGETHER with the current system. The data is all there and you must admit that many of the good players show up in the top 10. In the long run the current system may show to be better because it takes account of extra factors but on the short term it has its peculiarities so why not run the two in parallel for a while?

And again, we all know that the only thing that counts is winning a number of games (winning one game does not necessarily mean that you deal with a good player either and even with more games it depends who you play as you pointed out in your message). Intermediate results are fun but some of us will probably win the game without the power graph showing so untill turn 150 or so.

As for me, I am in the ranking for the fun of it and I realise that whatever we come up with it will not be 100% watertight (A "good" system would require many games over a relatively short time which is not feasible with PBEM-CTP). So, if it is decided not to show the extra view that that's OK with me.

[Keygen]

OK, no new topic...

What about wonders?
If I start building wonders after wonders in all of my games I will probably end up having the highest ratings of all!
Have you observed how high your power graph jumps when you complete a wonder? Well, a couple of turns later actually...

I don't suggest any change to the rating system. It might have some problems but it's there for quite some time and it shouldn't change at all.
But I suggest to point out all the problems that might have rised to possibly use this knowledge in some other rating system in some other future game, similar to CTP. It could also be helpful and fair for the current rated and future rated players to know among the possitive points of the ratings the negative points too so each one to have a better view of the system and be able to make his desicion easier and not dropping from them after.

[blackice]

"1. Blackice is incorrect about his conclusions regarding unfairly dropping in ratings. I could go on and on with example after example, but he is just mistaken. Trust me."

Teamhorses 4th place 04-28-2001

Starting points 21.723

BLACKICE 21.342
BLACKICE 20.966
BLACKICE 20.542
BLACKICE 20.679 down 1.044

High Voltage 1st place 04-25-2001

Starting points 21.364

BLACKICE 21.401
BLACKICE 21.630
BLACKICE 21.685
BLACKICE 21.723 up .359

Strife 1 st place 04-23-2001

Starting points 20.945

BLACKICE 20.945
BLACKICE 20.991
BLACKICE 21.094
BLACKICE 21.341
BLACKICE 21.364 up .419

Previous rating 008@ 20.995
Current rating 012@ 20.710


Quinns said "-You are quite right. The Rating shown is the
rating BEFORE adjustment. So your last
adjustment was with the 20.679 which then went to
20.710 upon defeating Darth Viper. Your rating
didn't go up that much in your first two games
because of the average lower level of players in
those games. But your recent losses in
Teamhorses set you back because you lost to lower
rated players (me for one )."

Sorry Quinns remember you posted it first I did not personally want to open this can of worms I am sure I made that clear?

I don't want to take all this thread up with this I believe Keygen has it right start a new one.

Now please if this win, win, loss continues, explain to me how it can be an incorrect conclusion? Givin the information above?

Keygen has it right with the wonders, it soars right Quinns So if he sat back and built wonders he would get points every ten turns. If he was the last of five players to lose to the winner, in the end the amount of points gained by him for the entire game vs the amount of points for the player who won the game would be huge, right?

Lets say the player that won was playing in five other games and just before he won he passes Keygen in the ratings. He then would be awarded little points for the game win. It would not compare to the ten turn points Keygen would make throughout a 200+ turn game. victory? Especially if the other players in the game were playing other games and were ahead of Keygen in the rankings throughout the game until the end. Cuz lets say he defeats them in this game but loses in the end to the higher ranked player.


How about an email of examples based on what you see above and please correct me if I am wrong.

This point system was based on a tennis ranking system right? Ok what tennis player gets ranked at 5th place over all without beating a single ranked player?

jpww no to your question, I think you were refering to my .5 point loss for inactivity while I was posted with a ranking and waiting for the first game I played to start? Right?

So enough said here lets start a new thread on it...

[quinns]

Blackice, you didn't post the level of players of the games you are winning in. Those wins are against low rated players. We have been through all this before. Let me use a exagerated example to make the point.

Suppose Blackice, rated 21 , plays the following players in 3 different games:

Game 1 --
Bill, rating 10
Joe, rating 8
Sam, rating 6

Game 2 --
Phillip 12
Michael 10
Ernesto 7

Game 3 --
Alfred 21
Bjorn 19
Charlie 18

Blackice wins the first two games hands down, and loses the third game in last place. According to your argument, your rating should go UP for winning the first two games and losing the third game. This is not at all fair. Am I the only one who sees it this way? Look at the LEVEL OF THE PLAYERS DEFEATED IN THE FIRST TWO GAMES.

[TheBirdMan]

Forget about the calculation I promised yesterday

Corse now I have read the background and the formulas for the ratingsystem.

My opinion is now, that one thing in Quinns calculations is wrong in all the games.

As this is "one" game I think all the calculation for each player should take start in the position, that this player had to each of the others and not as we do now.

This is how a game was calculated:
04-28-2001 T_HORSES 39 QUINNS 19.456 ZOBOZEWAR 20.000 0.282
04-28-2001 T_HORSES 39 QUINNS 19.738 FRANSES 19.240 0.222
04-28-2001 T_HORSES 39 QUINNS 19.960 BLACKICE 21.746 0.404
04-28-2001 T_HORSES 39 QUINNS 20.364 NUKE_BOY 19.962 0.231
04-28-2001 T_HORSES 39 QUINNS 20.595 DARTH_VIP 17.040 0.034
04-28-2001 T_HORSES 39 ZOBOZEWAR 19.718 FRANSES 19.018 0.201
04-28-2001 T_HORSES 39 ZOBOZEWAR 19.919 BLACKICE 21.342 0.376
04-28-2001 T_HORSES 39 ZOBOZEWAR 20.295 NUKE_BOY 19.731 0.216
04-28-2001 T_HORSES 39 ZOBOZEWAR 20.511 DARTH_VIP 17.006 0.035
04-28-2001 T_HORSES 39 FRANSES 18.817 BLACKICE 20.966 0.425
04-28-2001 T_HORSES 39 FRANSES 19.241 NUKE_BOY 19.515 0.259
04-28-2001 T_HORSES 39 FRANSES 19.500 DARTH_VIP 16.971 0.060
04-28-2001 T_HORSES 39 BLACKICE 20.542 NUKE_BOY 19.256 0.137
04-28-2001 T_HORSES 39 BLACKICE 20.679 DARTH_VIP 16.911 0.031
04-28-2001 T_HORSES 39 NUKE_BOY 19.119 DARTH_VIP 16.880 0.071

And this is how I think it should be calculated:
04-28-2001 T_HORSES 39 QUINNS 19.456 ZOBOZEWAR 20.000 0.282
04-28-2001 T_HORSES 39 QUINNS 19.456 FRANSES 19.240 0.244
04-28-2001 T_HORSES 39 QUINNS 19.456 BLACKICE 21.746 0.431
04-28-2001 T_HORSES 39 QUINNS 19.456 NUKE_BOY 19.962 0.278
04-28-2001 T_HORSES 39 QUINNS 19.456 DARTH_VIP 17.040 0.064
04-28-2001 T_HORSES 39 ZOBOZEWAR 20.000 FRANSES 19.240 0.194
04-28-2001 T_HORSES 39 ZOBOZEWAR 20.000 BLACKICE 21.746 0.401
04-28-2001 T_HORSES 39 ZOBOZEWAR 20.000 NUKE_BOY 19.962 0.250
04-28-2001 T_HORSES 39 ZOBOZEWAR 20.000 DARTH_VIP 17.040 0.046
04-28-2001 T_HORSES 39 FRANSES 19.240 BLACKICE 21.746 0.440
04-28-2001 T_HORSES 39 FRANSES 19.240 NUKE_BOY 19.962 0.302
04-28-2001 T_HORSES 39 FRANSES 19.240 DARTH_VIP 17.040 0.073
04-28-2001 T_HORSES 39 BLACKICE 21.746 NUKE_BOY 19.962 0.096
04-28-2001 T_HORSES 39 BLACKICE 21.746 DARTH_VIP 17.040 0.021
04-28-2001 T_HORSES 39 NUKE_BOY 19.962 DARTH_VIP 17.040 0.047

See, there are some differences. It really hurts when beaten by a lower ranked player. In fact that much, that turn to an unrated position in the run are much "cheaper" than stay ranked in a "bad" game if you are in the upper half of the ranks.

and then of course, we miss a colum or two last showing the actual rating points for the players.

Edit: And this would be their ratingpoints:

QUINNS 20,755
ZOBOZEWAR 20,609
FRANSES 19,617
BLACKICE 20,591
NUKE_BOY 19,083
DARTH_VIP 16,789


Edit-edit: How difficult can it be - more corrections to my calculations.

[This message has been edited by TheBirdMan (edited May 03, 2001).]
[This message has been edited by TheBirdMan (edited May 03, 2001).]
[This message has been edited by TheBirdMan (edited May 03, 2001).]
[This message has been edited by TheBirdMan (edited May 03, 2001).]